4.1 Sandwich core
55
Results
The simulation results are presented separately for the single layer and multi layer approaches. The correlation of simulation and experiments is benchmarked in terms of
stress-strain relationship for all four considered load cases. This correlation is illustrated
using three plots, where compression and tension in T-direction are combined in one
plot and the shear directions are each given in separate graphs. The experimental results
are plotted along with the simulation results. All obtained experimental curves are plotted (test scatter), while an average curve is highlighted. The simulations were carried out
on a six core Intel Xeon X5680 workstation. Computational effort varies greatly between
the different modelling approaches. Considering the compression load case, computation time is as low half an hour in case of the single layer approaches, while the multi
layer corner approach requires more than six hours.
SL approaches
The simulation results in comparison to the experiments for both single layer approaches
are given in Figure 44. The simulations generally approximate the curve progression obtained from the tests in all loading conditions. However, the sudden load drop shortly
after maximum stress in the compression and tension tests is not captured by the simulations. It is assumed that this is due to the implemented perfectly plastic material
model, which neglects brittle resin damage. The isotropic material model is established
as not capable to match all loading conditions simultaneously. In the present work the
material model is calibrated according to the compression experiments, which results in
a good approximation of this one load case. However, at the same time the strength of
all remaining load cases is overestimated. For shear loading the deviation is up to 15 %,
which can be considered acceptable for preliminary or rough predictions. However, in
case of tension the isotropic model exceeds the experimental curves by about 50 %. The
orthotropic material model enables more freedom when calibrating the material model.
Therefore, a good match of simulation and experiment is achieved for all loading conditions. However, there is one exception. The plateau stress under LT-shear is noticeably
overestimated by the orthotropic material model (Figure 44 bottom left). This effect is
attributed to the applied symmetry boundary conditions, which restricts the buckling
and folding of the cell walls located in the symmetry planes. Additionally performed
studies indicate that a larger scale along with applying a single symmetry plane would
result in a better approximation of the plateau stress when using the same material parameters. Another noticeable effect is the underestimated plateau strength under compression in case of the orthotropic model, whereas the isotropic model matches the plateau of the experiments well in case of the investigated honeycomb material. This may
be explained by the reduced tensile strength of the orthotropic model, which affects the
cell wall folding after buckling initiation, resulting in an increased drop in macroscopic
55
Results
The simulation results are presented separately for the single layer and multi layer approaches. The correlation of simulation and experiments is benchmarked in terms of
stress-strain relationship for all four considered load cases. This correlation is illustrated
using three plots, where compression and tension in T-direction are combined in one
plot and the shear directions are each given in separate graphs. The experimental results
are plotted along with the simulation results. All obtained experimental curves are plotted (test scatter), while an average curve is highlighted. The simulations were carried out
on a six core Intel Xeon X5680 workstation. Computational effort varies greatly between
the different modelling approaches. Considering the compression load case, computation time is as low half an hour in case of the single layer approaches, while the multi
layer corner approach requires more than six hours.
SL approaches
The simulation results in comparison to the experiments for both single layer approaches
are given in Figure 44. The simulations generally approximate the curve progression obtained from the tests in all loading conditions. However, the sudden load drop shortly
after maximum stress in the compression and tension tests is not captured by the simulations. It is assumed that this is due to the implemented perfectly plastic material
model, which neglects brittle resin damage. The isotropic material model is established
as not capable to match all loading conditions simultaneously. In the present work the
material model is calibrated according to the compression experiments, which results in
a good approximation of this one load case. However, at the same time the strength of
all remaining load cases is overestimated. For shear loading the deviation is up to 15 %,
which can be considered acceptable for preliminary or rough predictions. However, in
case of tension the isotropic model exceeds the experimental curves by about 50 %. The
orthotropic material model enables more freedom when calibrating the material model.
Therefore, a good match of simulation and experiment is achieved for all loading conditions. However, there is one exception. The plateau stress under LT-shear is noticeably
overestimated by the orthotropic material model (Figure 44 bottom left). This effect is
attributed to the applied symmetry boundary conditions, which restricts the buckling
and folding of the cell walls located in the symmetry planes. Additionally performed
studies indicate that a larger scale along with applying a single symmetry plane would
result in a better approximation of the plateau stress when using the same material parameters. Another noticeable effect is the underestimated plateau strength under compression in case of the orthotropic model, whereas the isotropic model matches the plateau of the experiments well in case of the investigated honeycomb material. This may
be explained by the reduced tensile strength of the orthotropic model, which affects the
cell wall folding after buckling initiation, resulting in an increased drop in macroscopic